linux-acpi.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: "Rafael J. Wysocki" <rjw@sisk.pl>
To: Len Brown <lenb@kernel.org>
Cc: Jeff Chua <jeff.chua.linux@gmail.com>,
	LKML <linux-kernel@vger.kernel.org>,
	ACPI Devel Maling List <linux-acpi@vger.kernel.org>,
	Linux-pm mailing list <linux-pm@lists.linux-foundation.org>,
	Matthew Garrett <mjg59@srcf.ucam.org>
Subject: [PATCH 3/11] ACPI: Fix acpi_os_read_memory() and acpi_os_write_memory()
Date: Thu, 20 Jan 2011 12:30:09 +0100	[thread overview]
Message-ID: <201101201230.10048.rjw@sisk.pl> (raw)
In-Reply-To: <201101201226.41021.rjw@sisk.pl>

From: Rafael J. Wysocki <rjw@sisk.pl>

The functions acpi_os_read_memory() and acpi_os_write_memory() do
two wrong things.  First, they shouldn't call rcu_read_unlock()
before the looked up address is actually used for I/O, because in
that case the mapping it belongs to may be removed before the I/O
is done.  Second, they shouldn't use the physical address if
there's no mapping for it, because that may lead to problems (at
least for the "write" case it's like sending data to a random
physical address in the hope it's going to work).

Make the rcu_read_unlock() be called by these functions when the I/O
is complete and modify them to dump stack and return error code if
the requested physical address is not present in the ACPI iomaps.

Signed-off-by: Rafael J. Wysocki <rjw@sisk.pl>
---
 drivers/acpi/osl.c |   31 +++++++++++++------------------
 1 file changed, 13 insertions(+), 18 deletions(-)

Index: linux-2.6/drivers/acpi/osl.c
===================================================================
--- linux-2.6.orig/drivers/acpi/osl.c
+++ linux-2.6/drivers/acpi/osl.c
@@ -638,17 +638,17 @@ acpi_os_read_memory(acpi_physical_addres
 {
 	u32 dummy;
 	void __iomem *virt_addr;
-	int size = width / 8, unmap = 0;
+
+	if (!value)
+		value = &dummy;
 
 	rcu_read_lock();
-	virt_addr = acpi_map_vaddr_lookup(phys_addr, size);
-	rcu_read_unlock();
+	virt_addr = acpi_map_vaddr_lookup(phys_addr, width / 8);
 	if (!virt_addr) {
-		virt_addr = acpi_os_ioremap(phys_addr, size);
-		unmap = 1;
+		rcu_read_unlock();
+		WARN(true, "Address not mapped: %llx\n", phys_addr);
+		return AE_BAD_ADDRESS;
 	}
-	if (!value)
-		value = &dummy;
 
 	switch (width) {
 	case 8:
@@ -663,9 +663,7 @@ acpi_os_read_memory(acpi_physical_addres
 	default:
 		BUG();
 	}
-
-	if (unmap)
-		iounmap(virt_addr);
+	rcu_read_unlock();
 
 	return AE_OK;
 }
@@ -674,14 +672,13 @@ acpi_status
 acpi_os_write_memory(acpi_physical_address phys_addr, u32 value, u32 width)
 {
 	void __iomem *virt_addr;
-	int size = width / 8, unmap = 0;
 
 	rcu_read_lock();
-	virt_addr = acpi_map_vaddr_lookup(phys_addr, size);
-	rcu_read_unlock();
+	virt_addr = acpi_map_vaddr_lookup(phys_addr, width / 8);
 	if (!virt_addr) {
-		virt_addr = acpi_os_ioremap(phys_addr, size);
-		unmap = 1;
+		rcu_read_unlock();
+		WARN(true, "Address not mapped: %llx\n", phys_addr);
+		return AE_BAD_ADDRESS;
 	}
 
 	switch (width) {
@@ -697,9 +694,7 @@ acpi_os_write_memory(acpi_physical_addre
 	default:
 		BUG();
 	}
-
-	if (unmap)
-		iounmap(virt_addr);
+	rcu_read_unlock();
 
 	return AE_OK;
 }

  parent reply	other threads:[~2011-01-20 11:30 UTC|newest]

Thread overview: 46+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-01-20 11:26 [PATCH 0/11] ACPI: Fixes and cleanups related to iomaps management Rafael J. Wysocki
2011-01-20 11:27 ` [PATCH 1/11] ACPI: Introduce acpi_os_ioremap() Rafael J. Wysocki
2011-01-20 11:28 ` [PATCH 2/11] ACPI / PM: Call suspend_nvs_free() earlier during resume Rafael J. Wysocki
2011-01-20 11:30 ` Rafael J. Wysocki [this message]
2011-01-20 11:30 ` [PATCH 4/11] ACPI: Do not export functions that are only used in osl.c Rafael J. Wysocki
2011-01-20 11:31 ` [PATCH 5/11] ACPI: Change acpi_ioremap_lock into a mutex Rafael J. Wysocki
2011-01-20 11:32 ` [PATCH 6/11] ACPI: Avoid walking the list of iomaps in osl.c twice in a row Rafael J. Wysocki
2011-01-20 11:33 ` [PATCH 7/11] ACPI: Make acpi_os_map_memory() avoid creating unnecessary mappings Rafael J. Wysocki
2011-01-20 11:34 ` [PATCH 8/11] ACPI: Do not use krefs under a mutex in osl.c Rafael J. Wysocki
2011-01-20 11:35 ` [PATCH 9/11] ACPI: Introduce acpi_os_get_iomem() Rafael J. Wysocki
2011-01-20 11:36 ` [PATCH 10/11] ACPI / PM: Use existing ACPI iomaps for NVS save/restore Rafael J. Wysocki
2011-01-20 11:37 ` [PATCH 11/11] ACPI / PM: Make NVS save/restore code use slightly less memory Rafael J. Wysocki
2011-01-20 16:06 ` [PATCH 0/11] ACPI: Fixes and cleanups related to iomaps management Jeff Chua
2011-01-20 20:57   ` Rafael J. Wysocki
2011-01-20 21:46     ` Jeff Chua
2011-01-21  0:04       ` Rafael J. Wysocki
2011-01-21  2:51         ` Jeff Chua
2011-01-21 12:01           ` Rafael J. Wysocki
2011-01-21 21:06           ` Rafael J. Wysocki
2011-01-22  5:54             ` Jeff Chua
2011-01-22  5:58               ` Jeff Chua
2011-01-22  9:25                 ` Rafael J. Wysocki
2011-01-22 17:24                   ` Jeff Chua
2011-01-22 19:12                     ` Rafael J. Wysocki
2011-01-23  0:14                       ` Jeff Chua
2011-01-23 20:28                         ` Rafael J. Wysocki
2011-01-24  1:13                           ` Jeff Chua
2011-01-24 21:36                             ` Rafael J. Wysocki
2011-01-22  9:13               ` Rafael J. Wysocki
2011-01-23 18:20                 ` Henrique de Moraes Holschuh
2011-01-23 20:35                   ` Rafael J. Wysocki
2011-01-23 23:15                     ` Henrique de Moraes Holschuh
2011-01-24 21:37                       ` Rafael J. Wysocki
2011-01-24 23:25 ` [PATCH 0/8] ACPI: Fixes and cleanups related to iomaps management (v2) Rafael J. Wysocki
2011-01-24 23:26   ` [PATCH 1/8] ACPI: Fix acpi_os_read_memory() and acpi_os_write_memory() (v2) Rafael J. Wysocki
2011-01-24 23:27   ` [PATCH 2/8] ACPI: Do not export functions that are only used in osl.c Rafael J. Wysocki
2011-01-24 23:28   ` [PATCH 3/8] ACPI: Change acpi_ioremap_lock into a mutex Rafael J. Wysocki
2011-01-24 23:28   ` [PATCH 4/8] ACPI: Avoid walking the list of memory mappings in osl.c twice in a row Rafael J. Wysocki
2011-01-24 23:29   ` [PATCH 5/8] ACPI: Make acpi_os_map_memory() avoid creating unnecessary mappings Rafael J. Wysocki
2011-01-24 23:30   ` [PATCH 6/8] ACPI: Do not use krefs under a mutex in osl.c Rafael J. Wysocki
2011-01-24 23:30   ` [PATCH 7/8] ACPI: Introduce acpi_os_get_iomem() Rafael J. Wysocki
2011-01-24 23:32   ` [PATCH 8/8] ACPI / PM: Use existing ACPI iomaps for NVS save/restore (v2) Rafael J. Wysocki
2011-02-05 15:31     ` Jeff Chua
2011-02-05 18:20       ` Rafael J. Wysocki
2011-02-05 18:24       ` Rafael J. Wysocki
2011-02-05 18:51       ` Linus Torvalds

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=201101201230.10048.rjw@sisk.pl \
    --to=rjw@sisk.pl \
    --cc=jeff.chua.linux@gmail.com \
    --cc=lenb@kernel.org \
    --cc=linux-acpi@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=linux-pm@lists.linux-foundation.org \
    --cc=mjg59@srcf.ucam.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).